[EM] Summary--Unstated CC assumptions

[MIKE OSSIPOFF]

Before I start, let me say that I resent some people's implication that 
there's no significant difference between me and Russ, as regards behavioral 
level, off-topicness, and negative worth.

Yes, I've sometimes rubbed people's nose in it, but only as a last resort. I 
try to make a "measured response", starting out very polite, and then, after 
a certain amount of arrogant, ill-mannered repetition, I mildly rebuke the 
person. Have you heard about those alarm-clocks that get louder gradually 
when you don't wake up?

I do that because, as Russ says, EM is apparently "The Wild West".  (Russ 
then proves it by postsing messages consisting of nothing but personal 
attack, and always getting away with it). In the Old West, you had to take 
the law into your own hands. If you want the guideliness enforced, you 
enforce them yourself. That's what I try to do when I start rubbing 
someone's nose in it. It's done with a positive intention, a good purpose.  
Don't interpret it negatively.

That's completely different from someone who posts long messages entirely 
devoted to expressing that person's opinion of someone else, filled with 
false statements*  intended to support that opinioin. _That's_ what is 
really off-topic. I've never done that.

*such as Russ's several false statements about the Python BeatpathWinner 
program that I'd sent to him.

Now, back on topic:

Summary--Unstated CC assumptions:

Russ says that the academic voting system authors intend something that they 
don't say--that they aren't saying what they intend. Russ says that as if he 
thinks that it contradicts something that I've said. Actually it's something 
that I've been saying for years, and have said a few times since Russ has 
been on the list: Most voting system academics do a very poor job of saying 
what they mean. That isn't a new discovery of yours, Russ.

But Russ has appointed himself to speak for those who don't say what they 
mean, by reading into what they say things that the rest of us don't find. 
Presumably by using ESP to determine what an author means when he defines a 
votes-only CC.

I'll get to that in a minute, but, first, contrary to what Russ seems to 
think, not all academic authors write definitions votes-only. Some have 
defined CC in terms of preference, but without stipulating how preference 
constrains voting, resulting in unmeetable criteria. My preference criteria 
differ from those in having that stipulation. So, before anyone says there's 
no precedent for preference criteria, I remind them that there are such 
criteria, poorly and sloppily defined though they are. So preference 
criteria aren't really a new and heretical thing. It's just that mine don't 
share the ridiculous fault of previous ones.

My point in starting that paragraph was that Russ mis-spoke when he told us 
that the academics only write votes-only CC.

Now, about the ESP. If someone writes a votes-only CC, having an unstated 
intention that will keep nonrank methods from passing, here are two 
completely different possibilities:

1. He intends, but doesn't state, that the criterion doesn't apply to 
nonrank methods.

2. He intends, but doesn't state, that nonrank methods fail the criterion by 
fiat.

This is where Russ's ESP comes in, because he thereby has determined that 
it's always #2.

But is it? One way we can judge that is by counting the voCC definitions 
that explicitlylsay which their intention is.

Well, Russ recently posted Blake's voCC (votes-only CC) definition. Why, I 
don't know, because it doesn't support his claim. Blake said "Application: 
Rank-Balloting".

Now, "Application: Rank Balloting" isn't the same as "Requirement: 
Rank-balloting". Russ, you won't understand that, so just take my word for 
it. If Blake had said that rank balloting was a requirement of his voCC, 
that would mean that nonrank methods fail. But when Blake says that rank 
balloting is the application of his voCC, that means that his voCC doesn't 
apply to nonrank methods. This will be obvious to most of you, but Russ can 
take my word for it.

So Russ's own criterion example that he posted, Blake's voCC, contradicts 
Russ's claim. If the people who don't state their intention share Blake's 
intention, then their intention is that voCC doesn't apply to rank methods. 
Russ, that is _not_ the same as saying that nonrank methods fail the 
criterion. I spell that out like that because you've previously indicated 
that you don't understand that distinction.

Do some authors intend, without stating it, that nonrank methods by fiat 
fail their voCC? Probably some do. Many or most, in fact. But we can't tell 
that from what they say. Unless we have ESP as Russ does. If we go by what 
some voCC definers actually say, then we have Blake's definition, whose 
stated intention is that his voCC doesn't apply to nonrank methods.

Anyway, as I said, probably many authors intend for nonrank methods to fail 
their voCC. But the evidence that I showed indicates that there's no reason 
to be so sure that all of them do.

Apparently ESP can be fallible.

Russ again thinks that he's contradicting me when he says that the usual 
intent of CC is that nonrank methods fail it. And again, that's something 
that I've been saying for a long time. I've said it long before Russ joined 
EM, and I've said it since Russ joined EM. I said it a few days ago when 
tellling James that I defined my preference CC (pCC) as I do so that it will 
comply with what I consider to be the intent of CC.

So though I question  Russ's way of arguing  for his claim, by finding what 
isn't there, it's a claim that I've been making all along.

But, somewhere along the line, somehow, in Russ's mind, the intent that 
nonrank methods fail CC became the intent that nonrank methods fail CC for 
no reason other than by definition (whether stated or not).

So, by Russ's reasoning then, my pCC violates Russ's supposed intention for 
CC, because, instead of saying that nonrank methods fail by fiat, those 
nonrank methods fail _on their own_, because of being nonrank methods. They 
do that without my mentioning nonrank methods or rank methods.

The kind of criteria that Russ advocates say "Nonrank methods fail because I 
say so." Somehow that isn't as convincing as having a _uniform_ test for all 
methods, which methods pass or fail, on their own, not because Russ says so. 
Rank methods fail because of their resulting deficiencies, not just because 
Russ says rank methods fail.

With a pairwise-count method, if everyone does nothing other than vote all 
of their preferences, requiring no strategic decisions, the CW wins. But 
with a method that doesn't let you vote all of your preferences,  it isn't 
so easy. Not being able to vote all of your preferences, you have know which 
ones to vote at the expense of the others. That's the problem with Approval 
and Plurality, and that's exactly why Approval and Plurality fail pCC. You 
can vote your preferences to the extent allowed by the method, but that 
isn't enough to elect the CW, even if everyone does that. That's the problem 
with Plurality and Approval, and that's why they fail pCC.

Russ said:

If a particular method is not ordinal, it fails the Condorcet Criterion by 
definition. It's a no-brainer.

I reply:

Have people noticed that very often the conclusions that someone like Russ 
refers to as a no-brainer are actually particularly stupid conclusions?

Mike Ossipoff

_________________________________________________________________
Express yourself instantly with MSN Messenger! Download today - it's FREE! 
http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/